Observation of Coherent Undulator Radiation from Sub- Picosecond Electron Pulses * Abstract

Transcription

1 SLAC PUB 716 September 1995 Observation of Coherent Undulator Radiation from Sub- Picosecond Electron Pulses * David Bocek and Michael Hernandez Physics Department and Stanford Linear Accelerator Center, Stanford University, Stanford, California 9439, USA Pamela Kung, Hung-chi Lihn, Chitrlada Settakorn, and Helmut Wiedemann Applied Physics Department and Stanford Linear Accelerator Center, Stanford University, Stanford, California 9439, USA Abstract The generation and observation of high-power, coherent, farinfrared undulator radiation from sub-picosecond electron bunches at the SUNSHINE facility is reported. Coherent undulator radiation tunable from 5 to µm wavelength is demonstrated. Measurements of the energy (up to 1.7 mj per 1 µs macropulse), frequency spectrum, and spatial distribution of the radiation are reported. Apparent exponential growth of the radiated energy as a function of undulator length is observed. Presented at Micro Bunches: A Workshop on the Production, Measurement and Applications of Short Bunches of Electrons and Positrons in Linacs and Storage Rings Upton, New York September 8-3, 1995 * Work supported by Department of Energy contract DE AC3 76SF515.

2 INTRODUCTION The generation of relativistic, sub-picosecond electron pulses allows the direct production of high-power, coherent, narrow-band, far-infrared radiation by passing the electron beam through a magnetic undulator (1). This provides a reliable and easily tunable source of far-infrared (FIR) radiation for scientific applications, without the need for an optical cavity. The Stanford UNiversity SHort INtense Electron-source (SUNSHINE) is a facility producing sub-picosecond electron pulses for the generation of coherent FIR, as described elsewhere in these proceedings (). This paper reports the observation of coherent far-infrared undulator radiation at the SUNSHINE facility. Phase coherence in radiation, such as occurs from bunches shorter than the radiated wavelength, leads to an increase in radiated energy as given by the form factor (3) Ne 1 ikz j f( k)= e, (1) Ne j= 1 where the z j are the longitudinal positions of the particles and N e number of particles in the bunch. The coherent energy is then is the total ( ) = ( ) E k N f( k) E k, () coh e 1e where E 1e (k) is the energy radiated by a single particle. The total energy radiated by a single electron of energy γ mc traversing an undulator is (4) E 1e = Nq u 6ε e π Kγ, (3) λ u where N u is the number of undulator periods, λ u is the period length, and K is the undulator strength parameter. Typical parameters for the generation of coherent undulator radiation at SUNSHINE are N u = 6, K = 1, λ u =. 77m, and γ = 31, so Eq. (3) gives 1 E 1e = 1 J. If a microbunch of N e = 1 8 electrons can be created with a form factor f(k) near its maximum value of one, then the coherent enhancement from

3 Eq. () means 4µJ will be radiated, which is 8% of energy of the microbunch. The possibility of such radiation efficiency motivates the experimental investigation reported here. EXPERIMENTAL SETUP The SUNSHINE facility has been described in detail elsewhere (5,6). In brief, it consists of an S-band thermionic rf gun, alpha magnet with energy collimation, a linac (phased to give 16 Mev beam energy in this experiment), and a 6-period, linearly-polarized, permanent-magnet undulator. Figure 1 shows the setup. Mechanical actuators allow foils to be inserted into the beamline to generate coherent transition radiation for bunch-length measurement (7,8) and to reflect out the coherent undulator radiation. A Michelson interferometer (M1) measures the electron bunch length before the undulator. A bolometer after the undulator measures the undulator radiation energy and spatial distribution. Another Michelson interferometer (M) measures the frequency spectrum of the undulator radiation. Typical parameters of the electron beam as adjusted for this experiment appear in Table 1. The estimated energy spread of the micropulse is the full width allowed through the energy collimator, and includes the energy-time correlation (energy slew) used to produce sub-picosecond bunches. The emittance is estimated from previous measurements (9). All measurements of radiation in this experiment were made using roomtemperature pyroelectric bolometers (Molectron P1-65). This was possible primarily because of the high intensity of the coherent radiation, and additionally because the bolometer electronics integrated over the entire 1µs macropulse. Thus the radiation measurements represent the contributions of the entire macropulse. TABLE 1. Typical electron beam parameters giving coherent undulator radiation. Parameter Value Units Macropulse rate 1 Hz Macropulse duration 1 µs Microbunch rate 856 MHz Microbunch duration.8 ps full width Microbunch population 1 8 electrons Microbunch energy 16 Mev Estimated micropulse energy spread/slew 3 % full width Estimated normalized emittance π mm-mradian 3

4 Alpha magnet with energy slit.5 Mev e- beam Linac 16 Mev Thermionic-cathode rf gun Actuator Actuator Actuator 16 Mev m long, 6 period undulator FIR Michelson interferometer M1 measured electron bunch length. Bolometer measured radiated energy. Michelson interferometer M measured energy spectral density. FIGURE 1. Schematic of experimental setup..3 Bolometer voltage [V] mirror postion [mm] FIGURE. Example interferogram of 16µm radiation. 4

5 MEASUREMENTS AND DISCUSSION Measurements at µm The investigation of coherent undulator radition was begun by setting the undulator strength to its maximum value K=3.. (This limit was imposed by the one inch external diameter of the wiggler chamber.) This was done to generate radiation at long wavelengths, where the most coherent enhancement was expected. The electron beam was tuned to optimize the total radiated energy collected in a copper condensing cone. The result was a bolometer voltage of 1.7 volts. This corresponds to 1mJ per macropulse if the factory calibration is assumed for the bolometer (1.7 volt/mj with the electronics used). A measurement of the average power was done using a Scientech laser power meter (thermopile), which was itself calibrated by inputting a known amount of electrical power. This gave a bolometer calibration of 1.9V/mJ, implying.9mj/macropulse collected in the cone. The frequency spectrum of the radiation was measured in Michelson interferometer M. As the path length of one arm of the interferometer is varied the interferometer generates the interferogram of the radiation pulse. The interferogram is the autocorrelation offset by a constant; the power spectrum is the Fourier transform of the autocorrelation. Figure shows a measured interferogram indicating that coherent undulator radiation was being produced. The coherence of the radiation was verified in two ways. The cross correlation (1.5 cm path difference in M) of radiation from each micropulse with that from the neighboring micropulse was demonstrated, and the radiated energy as a function of beam current was measured. (The high energy scraper in the alpha magnet was used to reduce the beam current.) The radiated energy varied as the square of the charge, as shown in Figure 3. 5

6 -.5 log(v_bolo[v]) log(n ) FIGURE 3. Measured undulator radiation as a function of number of particles squared. The fit line of slope one shows what is expected for coherent radiation. The spatial distribution of the radiation was measured by installing the bolometer on a translation stage and recording the bolometer voltage as a function of position. For this measurement, the bolometer was located 1.7m from the center of the undulator. The measured spatial distribution is shown in Figure 4, and corresponds to a total energy of 1.7mJ per macropulse. There appears to be a sharply peaked central cone of radiation with a base of 1cm diameter surrounded by a more slowly varying background. The sharpness of the central spike is noteworthy, as the coherent enhancement was expected to apply to the longer wavelengths that appear off-axis in undulator radiation, giving a more slowly varying distribution than measured. The pattern goes to zero outside the aperture of the polyethylene vacuum window. The meaning of this distribution is still under investigation; the measurement may be affected by radiation reflected from the inner surface of the undulator chamber, diffraction from the foil used to reflect the light, the size of the electron beam, and perhaps other factors. The magnitude of radiation is larger than expected. When the bolometer was located at the peak, it detected 15µJ of radiated energy per macropulse. The bolometer has an active surface radius of.5mm. Since it was located only.7m from an undulator of length m, the angular acceptance depended on the location of the source point in the undulator. For example, the bolometer accepted.9 milliradian from the beginning of the undulator and 3.6 milliradians from the 6

7 .5. V_bolo [V] y [mm] x [mm] 1 3 FIGURE 4. Measured spatial distribution of undulator radiation 1.7m from the center of the undulator. downstream end. The average solid angle accepted by a bolometer of radius r located at distance D from the dowstream end of an undulator of length L is dω = πr, (4) DD ( + L) which gives steradian here. The energy per solid angle radiated in the forward direction from a single electron of energy γ (4), after integrating about the fundamental frequency k, is du dω 1e = αn hck γ u K ( 1+ K / ) [ JJ], (5) where JJ = J K K J + K 1 + K. (6) 41 / 41 / ( ) ( ) Putting in the numbers gives J/steradian radiated at µm by one electron. But using steradian acceptance and the coherent enhancement of Eq. () with a measured N e = 1. 1 for the macropulse gives an expected 7

8 8µJ multiplied by f(k); it requires f(k) = 4.5 to match the measurement. Yet by definition, f( k) 1. This discepancy is not yet understood. The bunch length was measured using Michelson interferometer M1, and was found to be 4µm if a rectangular bunch is assumed, giving a form factor of sin( Lb/ ) sin( / ) f( k) =. π λ = π4 Lb/ π λ π4 / =, (7) implying that the measured energy is two orders of magnitude greater than expected from the theory. The next direction of investigation was to adjust the undulator to achieve the shortest radiated wavelength measurable in M. The wavelength of the radiation as a function of undulator strength parameter appears in Figure 5. The wavelength follows the expected form λ λu K = 1 +, (8) γ with a fit γmc = 17 Mev agreeing well with the measured 16 ± Mev. Radiated wavelength [µm] K FIGURE 5. Radiated wavelength vs. undulator strength parameter, fit by the resonance equation. 8

9 Energy spectral density [arbitrary units] wavenumber [cm -1 ] FIGURE 6. Measured spectrum of 47µm undulator radiation. Measurements at 47µm With the undulator strength set to K=.6, radiation at 47µm (wavenumber 15cm -1 ) was generated. The power spectrum of the radiation is shown in Figure 6. Atmospheric absorption, primarily due to water vapor, causes the ragged appearance of the spectrum. The spectrum is somewhat wider than is expected from the 6-period undulator; this may be due to the large energy spread over the macropulse. (Because the rf gun starts emitting before the linac is filled with rf, the energy increases at the beginning of the macropulse; because of transient beamloading, the energy decreases at the end of the macropulse.) 17µJ forward energy was measured with a bolometer located.3m from the dowstream end of the undulator; the angular acceptance defined in Eq. (4) was steradian and N e = 8 1 for the macropulse. With sin( Lb/ ) f( k) =. π λ Lb/ 1 = 1 Lb/ π λ π λ π4 / 47 = 4, (9) Eq. () and Eq. (5) give an expected 1.8µJ of undulator radiation. Again, much more radiation had occurred than was expected. This led to an investigation of the 9

10 dependence of radiated energy on undulator length, in order to test for selfamplified emission, which would be expected to show exponential growth (1). Radiated Energy at 47µm vs. Undulator Length To measure the radiated energy as a function of undulator length, ferromagnetic plates were placed between the undulator magnets and the beam pipe. These plates reduced the magnetic field at the beam from 8 to about 3 G, shifting the resonant frequency far out of the coherent regime. To change the undulator length in this manner, the linac had to be turned off briefly to allow access. The radiated energy was measured in the Michelson interferometer M. As periods at the downstream end of undulator were covered, the effective source point of the radiation moved upstream; the effective acceptance of M is estimated by Eq. (4) to account for this effect. The results are presented in Figure (7), and show an apparent exponential growth. A more detailed comparison with one dimensional self-amplified spontaneous emission theory appears elsewhere (11). SUMMARY In summary, coherent undulator radiation from sub-picosecond electron bunches has been observed in the 5-µm range. The observations show Macropulse energy per solid angle [arb. units] Number of undulator periods FIGURE 7. Semilog plot of measured undulator radiation as a function of undulator length. The fit line illustrates the exponential growth. 1

11 properties such as a narrower angular distribution and higher forward power than can be explained by ordinary coherent enhancement; these properties are under investigation. A measurement of forward energy at 47 µm vs. undulator length showed apparent exponential growth of the radiated energy, which is a signature of self-amplified spontaneous emission, and similiar measurements are being attempted at longer wavelengths. The interpretation of these results provides an interesting area for further research. ACKNOWLEDGEMENTS The authors thank R. Bonifacio, R. Carr, S. Krinsky, and H. D. Nuhn for useful discussions, and J. Sebek, J. R. Troxel, and the administration and staff of SSRL and HEPL for technical support. This research has been funded through SSRL and SLAC by D.O.E. Basic Energy Science Contract No. DE-AC3-76SF515. REFERENCES 1. H. Motz, J. Appl. Phys., 57 (1951).. H. Wiedemann, Observation of Coherent Radiation from Sub-picosecond Electron Bunches, Proceedings of this Workshop. 3. J. S. Nodvic and D. S. Saxon, Phys. Rev. 96, 18 (1954). 4. See, for eaxample, S. Krinsky, IEEE Trans. Nucl. Sci. NS-3, 378 (1983). 5. H. Wiedemann, P. Kung and H.-C. Lihn, Nucl. lnst. and Meth. A A319, 1 (199). 6. P. Kung, H.-C. Lihn, D. Bocek and H. Wiedemann, High-intensity Coherent FIR Radiation From Sub-picosecond Electron Bunches, in Gas, Metal Vapor, and Free- Electron Lasers and Applications, SPIE Conf. Proc. 118, 191 (1994). 7. H.-C. Lihn, D. Bocek, P. Kung, C. Settakorn and H. Wiedemann, SLAC-PUB and submitted to Phys. Rev. E. 8. P. Kung, H.-C. Lihn, D. Bocek and H. Wiedemann, Phys. Rev. Lett. 73, 967 (1994). 9. M. Borland, A High-Brightness Thermionic Microwave Electron Gun, SLAC-Report-4, (1991). 1. See, for example, R. Bonifacio, C. Pellegrini and L. M. Narducci, Opt. Commun. 5, 373 (1984). 11. D. Bocek, P. Kung, H.-C. Lihn, C. Settakorn and H. Wiedemann, SLAC-PUB and submitted to Phys. Rev. Lett. 11